langchain
langchain copied to clipboard
langchain-ai
[bugfix] Fix persisted chromadb vectorstore
If a persist_directory param was set, chromadb would throw a warning that ""No embedding_function provided, using default embedding function: SentenceTransformerEmbeddingFunction". and would error with a Illegal instruction: 4 error.
This is on a MBP M1 13.2.1, python 3.9.
I'm not entirely sure why that error happened, but when using get_or_create_collection instead of list_collection on our end, the error and warning goes away and chroma works as expected.
Added bonus this is cleaner and likely more efficient. list_collections builds a new Collection instance for each collect, then Chroma would just use the name field to tell if the collection existed.
@timothyasp theres been lots of dev on chroma side - is this error still happening with recent versions?
I'm seeing this right now with chromadb==0.3.11 and langchain==0.0.105
So yes, I think it's still happening. And I'm using the persist_directory parameter.
Yea, I believe so, I’ve been using this patched code now and been pulling pretty much daily from both upstreams, and things have been working.
On Wed, Mar 8, 2023 at 8:17 PM Harrison Chase @.***> wrote:
@timothyasp https://github.com/timothyasp theres been lots of dev on chroma side - is this error still happening with recent versions?
— Reply to this email directly, view it on GitHub https://github.com/hwchase17/langchain/pull/1444#issuecomment-1461255453, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAFMY4ZX62S45XEIVH6VZZTW3FKV5ANCNFSM6AAAAAAVP57MYA . You are receiving this because you were mentioned.Message ID: @.***>
-- -Tim
tried the fix looks good to me looking forward for a merge in a new release please, this issue is annoying
@hwchase17 can we get this merged? It seems to be impacting a lot of people
In the meantime I forked the original repos and applied the patch (to be up to date with latest commit).
You can switch your langchain by langchain@git+https://github.com/lambda-science/langchain.git in a simple requirements.txt.
Not optimal at all for now, but it does the job
Do we know if this error has been resolved I am getting an error message as follows: "You must provide embeddings or a function to compute them". How do I patch my current vectorstore chromadb file?
